ORI GIN AL PA PER

Towards a Validation of the Three Pathways Modelof Pathological Gambling

Marc Valleur1• Irene Codina1

• Jean-Luc Venisse2,3•

Lucia Romo4,5,6• David Magalon7

• Melina Fatseas8•

Isabelle Chereau-Boudet9• Mohamed-Ali Gorsane10,11

•

Alice Guilleux3,12• Groupe JEU • Marie Grall-Bronnec2,3

•

Gaelle Challet-Bouju2,3

� Springer Science+Business Media New York 2015

Abstract With the aim of validating the three pathways hypothesis of pathological

gambling (Blaszczynski and Nower in Addiction 97:487–499, 2002) 372 pathological

gamblers meeting DSM IV (2000) criteria were assessed via a structured clinical interview

as well as being subjected to personality tests and evaluation of their gambling practices.

Our results show that it is possible to identify three subgroups corresponding to the three

pathways: behaviourally conditioned problem gamblers, emotionally vulnerable problem

gamblers and antisocial impulsivist problem gamblers. Our results particularly demonstrate

that impulsivist gamblers preferentially choose semi-skilful gambling (horse racing and

sports gambling) whereas emotionally vulnerable gamblers are significantly more attracted

to games of chance (one-armed bandits, scratch cards, etc.) This led us to propose a

functional presentation of the three pathways model which differs somewhat from the

Blaszczynski and Nower presentation.

& Marc Valleurmarc.valleur@gmail.com

1 Marmottan Medical Center, GPS Perray-Vaucluse, Hopital Marmottan, 17 rue d’Armaille,75017 Paris, France

2 Clinical Investigation Unit BALANCED ‘‘BehaviorAL AddictioNs and ComplEx moodDisorders’’, Department of Addictology and Psychiatry, University Hospital of Nantes, Nantes,France

3 EA 4275 SPHERE ‘‘bioStatistics, Pharmacoepidemiology and Human sciEnces Research tEam’’,Faculties of Medicine and Pharmaceutical Sciences, University of Nantes, Nantes, France

4 EA 4430 CLIPSYD «CLInique PSYchanalyse Developpement», University of Paris OuestNanterre La Defense, Nanterre, France

5 Assistance Publique – Hopitaux de Paris (APHP), Louis Mourier Hospital of Colombes, Colombes,France

6 Present Address: Psychotherapies Unit, Sainte-Anne Hospital – Psychiatry and Neurosciences,Paris, France

7 Department of Adult Psychiatry, Sainte-Marguerite University Hospital of Marseille, Marseille,France

123

J Gambl StudDOI 10.1007/s10899-015-9545-y

Keywords Problem gambling � Etiology � Gambling subtypes � Co morbidity �Impulsivity � Pathway model � Addiction

Introduction

The recent classification of pathological gambling in the nosographic category of ‘‘sub-

stance-related and addictive disorders’’ (D.S.MV 2013) is new and marks a return to a very

old tradition: the passion for gambling was considered to be a disease in the sixteenth

century (Nadeau and Valleur 2014), after amorous passion and before drunkenness and the

various drug addictions. According to this ancient point of view addictions ‘‘with drugs’’

may only form a particular subgroup of these invasive passions, each of which asks both

the question of the specific ‘‘toxicity’’ of the subject of addiction and that of individual

vulnerability. Concerning gambling, factors concerning individuals are very important,

since only a small proportion of ‘‘consumers’’ are affected, whereas it is a widespread

activity in society as a whole.

Studies of the prevalence of pathological gambling in the general population certainly

show a certain disparity according to country, from 0.2 % in Norway to 2.1 % in Australia

(INSERM 2008; Shaffer and Hall 2001). But generally speaking, in spite of cultural

differences, gaming offers and terms of regulation, this disparity is quite relative: in most

European countries, its prevalence is around 0.5 % of the adult population (mean preva-

lence rate 0.6 %, from 28 studies analysed by Planzer et al. 2014). France is therefore

within the average, with 0.4 % of pathological gamblers and 0.9 % of problem gamblers

(Costes et al 2011). Despite great differences in gambling offers and regulation, France and

United Kingdom are for instance very close in terms of pathological gambling prevalence:

in U.K., rate was 0.5 % in 2000 (Sproston et al.), 0.4 % in 2007 (Wardle et al.), 0.55 % in

2011 (Wardle et al.). Big changes in gambling offers in Sweden did not lead to long lasting

changes in pathological gambling prevalence, and Binde (2013) describes this fact as a

«homeostatic system».

This relative stability in prevalence is at least partly due to the fact that some people are

more inclined than others to become addicts, and leads to highlighting not only regulatory

methods and the addictiveness of different games, but also individual vulnerability factors

and motivations not only to gamble, but to gamble excessively. It is therefore very im-

portant to know the various pathways which lead someone to dependence on gambling.

In an article published in 2002, Alex Blaszczynski and Lia Nower suggested separating

pathological gamblers into three groups, depending on the pathway involved (Blaszczynski

and Nower 2002):

8 Psychiatry Laboratory, Sanpsy CNRS USR 3413, University of Bordeaux and Charles PerrensHospital, Bordeaux, France

9 Psychiatry Department, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France

10 Psychiatry and Addictology Department, Assistance Publique – Hopitaux de Paris (APHP), PaulBrousse University Hospital of Villejuif, Villejuif, France

11 Present Address: Henri EY Hospital, GPS Perray-Vaucluse, Paris, France

12 Unit of Methodology and Biostatistics, University Hospital of Nantes, Nantes, France

J Gambl Stud

123

• ‘‘Behaviourally conditioned’’ gamblers (Group 1), who gamble for family, cultural or

subcultural reasons, and because of their gambling history itself.

• ‘‘Emotionally vulnerable’’ gamblers (Group 2), who have particular vulnerability

factors as well as gambling conditioning: anxiety or depression, negative life events,

poor coping strategies.

• ‘‘Antisocial, impulsivist’’ gamblers (Group 3), who add impulsiveness and antisocial

type personality disorders to the previous group’s factors.

The authors therefore provide a classification in three groups, which differs particularly

in the associated comorbidities: ‘‘behavioural conditioning’’ is common to all pathological

gamblers and group 1 therefore comprises ‘‘pure’’ gamblers. Group 2 adds emotional

vulnerability to this conditioning. Group 3 has the same characteristics as group 2, but with

an additional impulsiveness factor, sometimes antisocial in nature.

A number of articles (in Pubmed, the Nower and Blaszczynski article is cited in 261

articles; in Google Scholar, in 489 articles) referring to this classification as a fact, no doubt

because it corresponds to elements which are regularly observed clinically (Bonnaire,

2011; Martins et al. 2011; Ledgerwood and Petry 2010; Milosevic and Ledgerwood

2010…). Milosevic and Ledgerwood (2010) have inventoried all the literature concerning

the classification of pathological gamblers into separate subgroups. Their analysis shows

that, from the first study by Moran (1970), to the recent publication by Ledgerwood and

Petry (2010), there is a coherence, compatible with the Blaszczynski and Nower model.

By studying all this literature, it is indeed possible to distinguish three groups:

• The first group is classified as subcultural (Moran 1970), normal (Lesieur 2001),

conditioned (Blaszczynski and Nower 2002; Ledgerwood and Petry 2010), social

(Stewart and Zack 2008), simple gambler (Vachon and Bagby 2009). This group is the

Blaszczynski and Nower group 1, ‘‘Behaviourally conditioned’’.

• The second group is called neurotic (Moran 1970), depressive or anxious (Graham and

Lowenfeld 1986), repetitively depressive (McCormick 1987), tending to depression

(Blaszczynski et al. 1990), seeking escape (Lesieur and Blume 1991; Ledgerwood and

Petry 2006), psychologically anxious (Steel and Blaszczynski 1996), emotionally

vulnerable (Blaszczynski and Nower 2002; Ledgerwood and Petry 2010), confronting

situations (Stewart and Zack 2008). This group therefore meets the criteria for

Blaszczynski and Nower’s group 2.

• The third group has been classified, according to authors, as impulsive (Moran 1970),

passive-aggressive or emotionally unstable (Graham and Lowenfeld 1986), chronically

under-stimulated (McCormick 1987), tending to boredom (Blaszczynski et al. 1990),

action seeker (Lesieur 2001; Lesieur and Blume 1991), anti-social impulsive (Steel and

Blaszczynski 1996), impulsive anti-social (Blaszcsynski and Nower 2002; Ledgerwood

and Petry 2010)… This group thus corresponds to Blaszczynski and Nower group 3.

Attempts to validate these groups however, according to the meta-analysis of Milosevic

and Ledgerwood (2010) have not been quite conclusive to date. Our research thus aims to

verify the validity of this typology in three groups for a large cohort of problematic

gamblers. Indeed, we assume that groups can be constituted on the basis of variables

identified as the most ‘‘typical’’ of the three groups (e.g. the presence of anxiety or mood

disorders prior to the gambling problem, for the ‘‘emotionally vulnerable’’ group, the

presence of an antisocial personality disorder or a high level of novelty seeking for the

‘‘antisocial impulsive’’ group, the others forming the ‘‘behaviourally conditioned’’ group,

and that these groups are close to the description given by Blaszczynski and Nower (2002).

J Gambl Stud

123

We would also like to go beyond this model and describe the three types more functionally.

Our main hypothesis is that:

• Blasczsynski’s group 2 subjects (emotionally vulnerable) present more anxiety-

depressive disorders, even apart from those preceding the gambling problem. It is also

expected that this group will contain proportionally more women and elderly people,

that the age at which gambling started is higher and that they gamble more often to

escape from or relieve problems.

• Group 3 subjects (antisocial impulsivists) lie more often to hide the extent of their

gambling and more often endanger their personal relationships or their work. We also

expect to find more men and younger people in this group, who began gambling early

in life. We would also expect to find evidence of other addictive behaviour.

• Group 1 subjects (‘‘behaviourally conditioned’’) show intermediate scores for most of

the dimensions evaluated. According to some authors (Turner et al. 2008), we may

expect this group to have higher scores for gambling-related cognitions. Moreover,

these subjects are more likely to have a family history of excessive gambling.

We add to this criteria the fact that the type of gambling involved are supposed to differ

between groups. Indeed, the different types of gambling match particular motivations. The

gambling opportunities available on the market include games of pure chance (lottery,

scratch cards, one-armed bandits, roulette, video lottery machines…), games of chance

requiring some skill (sports betting, horse racing, blackjack) and skilful games of chance

(poker), depending on whether the game involves some skill and is played against an

operator or other players (Boutin 2010). We have also highlighted the strict opposition

between ‘‘dream games’’ such as big lotteries, which are not very addictive, and

‘‘excitement games’’ such as one-armed bandits or horse racing, which often lead to

gambling problems (Sharpe 2002; Valleur and Bucher 2006; Bonnaire et al. 2009). It

would seem logical that ‘‘antisocial impulsivist’’ gamblers prefer games with a competitive

component (Caillois et al. 1957) and thus mainly invest in skilful or semi-skilful games

where tension is high and the feeling of mastering or controlling the game is very high.

‘‘Behaviourally conditioned’’ gamblers are more likely to go for games with a high illusion

component: ‘‘Rapido’’, one-armed bandits, in fact games of pure chance, but in which the

illusion of control is very present. Emotionally vulnerable gamblers are more likely to

agree to play games of pure chance, choosing ‘‘hypnotic’’ or ‘‘anaesthetic’’ games pro-

moting dissociation as described by Jacobs (1988), such as one-armed bandits or lotteries

which promote dreams and escape.

Methods

Participants

The participants were 372 problem gamblers who took part in the JEU cohort study that is

currently taking place. The JEU study is a 5-year longitudinal case–control cohort per-

formed at national level. It includes, more broadly, 628 gamblers, divided into three

groups: non-problem gamblers (NPG), problem gamblers without treatment (PGWT) and

problem gamblers seeking treatment (PGST) (for more information, please refer to the

study protocol of the JEU cohort: Challet-Bouju et al. 2014). The present study was

performed on the baseline data of the JEU cohort and on problem gamblers only [that is

PGWT (n = 203) ? PGST (n = 169)]. PGWT were recruited in various gambling places

J Gambl Stud

123

(casinos, cafes, smoke shops, etc.) and via the press, in order to cover the broadest possible

range of gambling activities. PGST were recruited in seven care centres, among patients

who started treatment\6 months before. Only participants who reported gambling on at

least one occasion in the previous year and who were between 18 and 65 years old were

included in the study.

Assessment Procedure

The proposed assessment includes a clinical structured interview, carried out with a trained

researcher or psychologist, with a set of standardized self-report questionnaires. The

assessment comprised socio-demographic, gambling-related and clinical data, which are

described below.

Socio-Demographic Characteristics

A short questionnaire included a few questions about gender, age, marital status, profes-

sional activity, educational level and level of income.

Gambling Habits

A detailed interview was created to explore the gambling habits. Participants were asked

about their participation in various forms of gambling over the past year, their preferred

gambling activity (i.e. the gambling form which they preferred to play out of all the

gambling activities they had tried in their lives), monthly gambling expenditure especially

in relation to income, maximum wagering in a single day, the age at which they were

initiated into gambling and their family history of problem gambling.

Pathological Gambling Section on the DSM-IV (APA 2000)

Problem gamblers were identified through an interview based on the 10 diagnostic criteria

for PG in the DSM-IV. Gamblers who met at least three DSM-IV criteria were classified as

problem gamblers (including both gamblers ‘‘at risk’’ for pathological gambling and

gamblers with a diagnosis of PG). We used a non-standard threshold of three instead of five

to include subclinical forms of problem gambling, which could be considered as forms of

‘‘abuse of gambling’’ similar to the notion of substance abuse. Previous literature supported

the relevance of this categorization (Potenza 2006; Toneatto and Millar 2004).

South Oaks Gambling Scale (SOGS) (Lesieur and Blume 1987; Lejoyeux 1999)

The SOGS is a 20-item self-report questionnaire used to assess the severity of gambling

problems.

Gambling Attitudes and Beliefs Survey—Revised Version (GABS-23) (Breenand Zuckerman 1999; Bouju et al. 2014)

The GABS is a self-report questionnaire which assesses irrational beliefs and attitudes

about gambling. The GABS-23 is a revised version of the original GABS which consists of

23 items divided into 5 dimensions: Strategies, Chasing, Attitudes, Luck and Emotions.

J Gambl Stud

123

Mini International Neuropsychiatric Interview—Fifth Version (MINI) (Lecrubier et al.1997)

This short diagnostic structured interview explores the main axis-I psychiatric disorders

(plus current risk of suicide and antisocial personality disorder) of the DSM. It includes

assessment of the major anxiety disorders, mood disorders, addictive disorders and to a

lesser extent psychotic disorders. Once a diagnosis was validated, the age at which the first

episode occurred was compared with the age at which the gambling problem began, to

define whether the disorder existed before or after the gambling problems.

Wender-Utah Rating Scale-Child (WURS-C) (Ward et al. 1993; Bayle et al, 2003)

The WURS-C is a self-report questionnaire used in adults to make a retrospective

assessment of Attention Deficit/Hyperactivity Disorder (ADHD) in childhood. Authors

recommended using a threshold of 46/100 to make a retrospective screening of ADHD in

childhood.

Temperament and Character Inventory—125 (TCI-125) (Chakroun-Vinciguerra et al.2005; Cloninger 1992; Pelissolo and Lepine 2000)

The shorter 125-item version of the TCI is a self-report questionnaire used to rapidly

explore the seven dimensions of personality defined by Cloninger’s psychobiological

model (Cloninger et al. 1993). The TCI-125 assesses four temperament traits (Novelty

Seeking, Harm Avoidance, Reward Dependence and Persistence) and three character traits

(Self-Directedness, Cooperation and Self-Transcendence).

Constitution of Typical Groups

Three groups were set up based on the characteristics described by Blaszczynski and

Nower (Blaszczynski and Nower 2002). For group 2 (which we set up to resemble

Blaszczynski and Nower’s ‘‘emotionally vulnerable’’ group), we chose a single criterion:

the fact that the subject had suffered at least one episode of anxiety or depression before

having a gambling problem. For group 3 (which we set up to resemble Blaszczynski and

Nower’s ‘‘antisocial impulsivist’’ group), we isolated: (1) gamblers suffering from anti-

social personality disorder, (2) gamblers with a particularly high score for the TCI ‘‘novelty

seeking’’ component (i.e. for those in whom ‘‘novelty seeking’’ is predominant amongst the

other personality components). Finally for group 1 (which we set up to resemble

Blaszczynski and Nower’s ‘‘behaviourally conditioned’’ group), in the absence of obvious

classification criteria, we decided to group all the subjects who did not belong to group 2 or

group 3 in this group.

Statistical Analysis

The first stage of the analysis consisted of describing the three groups thus formed, notably to

check that apart from the variables used to set up the groups, they also displayed charac-

teristics matching the theoretical model described by (Blaszczynski and Nower 2002).

Then, since the pathway model is an evolutionary model (group 2 is equivalent to group

1 plus emotional and biological vulnerability, and group 3 is equivalent to group 2 plus

J Gambl Stud

123

impulsive traits), we conducted two analyses: the first one to study the differences between

group 1 and group 2, and the other one for the differences between group 2 and group 3.

(We also compared group 1 to group 3, but as this did not add new data to the results, we

do not present it here). We used logistic regression to compare the characteristics of the

groups. At first, bivariate analyses were performed by introducing the variables one by one.

Only variables which were significant at 20 % were then included in a multivariate logistic

regression. Non-significant variables at 5 % were removed one at a time, starting with the

least significant variable (backward procedure), in order to select only the variables which

provided significant information in the model.

Ethical Considerations

Participants gave their written informed consent. This study was approved by the French

Research Ethics Committee.

Results

The Existence of Three Separate Groups of Gamblers

Description of the Groups

Firstly, we would like to describe the three groups constituted, in order to make sure that

they resemble the theoretical typologies described by (Blaszczynski and Nower 2002). In

spite of the very limited criteria, we obtained three very distinct groups:

Group 1 (Supposed ‘‘Behaviourally conditioned’’) included 162 subjects (43.6 % of

the total sample). Men were over-represented (72.8 %) and the average age was 43.5

(sd = 12.6). Nearly half the gamblers in this group lived alone (43.8 %) and most had

an education level below ‘A’ level (55.3 %). Most were working (64.8 %) and had a

regular income at least equivalent to the minimum wage (74.7 %). Concerning

comorbidities, this group recorded 30.9 % of subjects with a history of mood disorders,

20.4 % a history of anxiety disorder, 30.3 % with an addictive disorder (other than

excessing gambling) and 3.1 % with a history of psychotic syndrome.

Group 2 (‘‘Emotionnally vulnerable’’) included 111 subjects (29.8 % of the total

sample). Men were overrepresented here also, but to a lesser extent than in group 1

(67.6 %). The average age was 44.1 (sd = 13.1). The gamblers in this group were more

likely to live alone (55.9 %) and they were almost as many to have a level of education

level below ‘A’ level (49.1 %) as higher or equal to it (50.9 %). Most gamblers in this

group were working (59.5 %) and had a regular income at least equal to the minimum

wage (68.5 %). Naturally, because of the group constitution criteria, most of the subjects

had a history of mood disorders (80.2 %) or anxiety disorders (74.8 %). Furthermore,

46.0 % had a history of addictive disorders (other than excessive gambling) and 20.7 %

a history of psychotic syndrome.

Finally, group 3 (‘‘Impulsivist’’) included 99 subjects (26.6 % of the total sample).

There was a distinct male predominance (83.8 %) and the average age was 41.8

(sd = 11.6). They mainly lived as a couple (55.1 %) and usually had an education level

below that of ‘A’ level (56.6 %). They were usually working (69.7 %) and had regular

incomes at least equal to the minimum wage (63.6 %). Concerning associated

J Gambl Stud

123

psychiatric comorbidities, 53.5 % of the gamblers in this group had a history of mood

disorders, 36.4 % a history of anxiety disorders, 50.5 % a history of addictive disorders

other than excessive gambling, and 9.1 % a history of psychotic syndrome.

Comparisons Between the Groups (Socio-Demographic, D.S.M. Items, Comorbidity,Personal History)

Comparison of Group 1 (Behaviourally Conditioned) and Group 2 (Emotionally Vul-

nerable): The results of the multivariate logistic regression comparing gamblers in group

2 (emotionally vulnerable) to gamblers in group 1 (behaviourally conditioned) are given in

Table 1. They show that group 2 gamblers are more likely to meet DSM criteria 4 (irri-

tation during attempts to reduce their gambling practice) and 8 (illegal acts to finance the

practice). They give a lower score to GABS-23 ‘‘strategies’’ and a lower score to the TCI-

125 ‘‘determination’’ dimension. They are more likely to display addictive or psychotic

disorders before developing the gambling problem, and more of them have a ‘‘moderate’’

risk of suicide.

Comparison of Group 2 (Emotionally Vulnerable) and Group 3 (Antisocial Impulsivist):

Proportion of the Antisocial-Impulsive Characteristic The results of the multivariate

logistic regression comparing group 3 gamblers (antisocial impulsivist) with those in group

2 (emotionally vulnerable) are given in Table 2. They show that group 3 gamblers are

more likely to meet DSM criterion 9 (endangering personal or work relationships). For

group 3 gamblers, the gambling problem was usually discovered by the family by accident.

They are therefore subjects who gamble in a solitary fashion, with very little socialisation

or sharing. They are more often found to have a ‘‘high’’ risk of suicide, which confirms a

marked trend towards an actual attempt, depression not being significantly higher than in

the other groups (depression: group 1: 29.81 %; group 2: 74.77 %; group 3: 46.46 %.

Finally they have lower scores for inclination to avoid danger, dependence and reward, and

persistence.

Comparisons Between the Groups (Choosing the Preferred Game)

As shown in Table 1, the multivariate logistic regression comparing group 1 gamblers with

group 2 shows that group 2 gamblers (emotionally vulnerable) are more likely to choose

games of pure chance as their favourites, compared with group 1 gamblers (behaviourally

conditioned). Although the other regression comparing group 2 to group 3 did not show

any evidence of difference in the game of choice between group 3 (antisocial impulsivists)

and group 2 (emotionally vulnerable) gamblers, the descriptive elements still seem to

confirm our hypotheses.

Indeed, if you look at the proportions of gamblers playing the three main types of game

(pure chance, semi-skilful and skilful) depending on the group to which the gambler

belongs, you can see that:

• 48.5 % of gamblers in group 3 (‘‘antisocial impulsivists’’) prefer semi-skilful games,

38.4 % pure chance games and finally 13.1 % for games requiring skill.

• On the other hand, 64.0 % of gamblers in group 2 (‘‘emotionally vulnerable’’) prefer

games of pure chance, 27.9 % semi-skilful games and finally 8.11 % of games

requiring skill. The proportions are thus reversed.

J Gambl Stud

123

• And group 1 (‘‘behaviourally conditioned’’) is intermediate between the other two, with

47.8 % preferring games of pure chance, 37.1 % for semi-skilful games and 15.1 % for

skilful games.

By detailing the games involved a little more, the differences are underlined more

clearly:

• Scratch cards (‘‘pure chance’’) were preferred by 13.8 % of group 2, versus only 7.2 %

in group 3 and 7.0 % in group 1.

• One-armed bandits (pure chance, with often illusion of control) were preferred by

37.6 % in group 2, versus 20.6 % in group 3 and 26.0 % in group 1.

• Instant lotteries were preferred by 4.1 % in group 3, versus 3.7 % in group 2 and 1.3 %

in group 1.

Table 1 Results of logistic regression comparing group 1 (behaviourally conditioned) and group 2(emotionally vulnerable): the contribution of emotional and biological vulnerability

Variables Group 1 %or M (sd)

Group 2 %or M (sd)

OR p

DSM criterion 4 (irritation) 38.9 % 61.3 % 3.04*** \0.001

DSM criterion 8 (illegal acts) 12.4 % 22.5 % 2.39* 0.044

Type of gambling1 0.008

Pure chance 47.8 % 64.0 % – –

Semi-skilful 37.1 % 27.9 % 0.47* 0.035

Skilful 15.1 % 8.1 % 0.21** 0.006

GABS ‘‘strategies’’ 48.0 (23.6) 43.0 (23.0) 0.98** 0.001

TCI ‘‘determination’’ 66.5 (18.0) 56.1 (20.0) 0.98** 0.007

Presence of an addictive disorder(other than excessive gambling)before the start of gambling problems

80.0 % 82.9 % 3.67** 0.002

Presence of a psychotic problembefore the start of gambling problems

100.0 % 81.0 % 17.08** 0.002

Level of risk of suicide2 0.043

Low 56.7 % 58.3 % – –

Moderate 3.3 % 22.9 % 14.66* 0.013

High 40.0 % 18.8 % 0.90 0.862

The ORs are given in reference to group 1, i.e. if the OR is greater than 1, the characteristic evaluated isfound more in group 2 than group 1

OR = odds ratio

% = proportions of the procedure studied within the group (qualitative variables)

M (sd) = mean and standard deviation of the variable studied within the group (quantitative variables)1 Games of pure chance are used as reference for the comparison of types of gambling (i.e. the figurescorresponding to ‘‘semi-skilful’’ reflect the comparison of ‘‘semi-skilful’’ versus ‘‘pure chance’’ and thosecorresponding to ‘‘skilful’’ reflect the comparison of ‘‘skilful’’ versus ‘‘pure chance’’2 The ‘‘low’’ level is used as reference for the comparison of levels of suicide risk (i.e. the figurescorresponding to ‘‘moderate’’ reflect the comparison of ‘‘moderate’’ versus ‘‘low’’ and those correspondingto ‘‘high’’ reflect the comparison of ‘‘high’’ versus ‘‘low’’)

* p\ 0.05, ** p\ 0.01, *** p\ 0.001

J Gambl Stud

123

• Roulette was preferred by 4.4 % in group 1, versus 4.1 % in group 3 and 3.7 % in

group 2.

• Draws were preferred by 8.9 % in group 1, versus 5.5 % in group 2 and 3.1 % in group

3.

• Horse race betting was preferred by 34.0 % in group 3, versus only 19.3 % in group 2

and 26.0 % in group 1.

• Sports betting was preferred by 12.4 % in group 3, versus 8.3 % in group 2 and 10.8 %

in group 1.

• Finally, poker (considered as a ‘‘skilful game’’) was preferred by 13.4 % in group 3,

versus 8.3 % in group 2 and 15.2 % in group 1.

Group 1 was usually situated in an intermediate position between the other two groups,

as predicted.

Discussion

From a purely descriptive point of view, the results of our study seem to indicate that our

group 2, defined by the existence of at least one episode of anxiety or depression before

gambling, is similar to Blaszczynski and Nower groups 2 (‘‘emotionally vulnerable’’).

Our group 3 (defined by ‘‘novelty seeking’’, or presence of A.P.D) is similar to

Blaszczynski & Nower group 3 (‘‘antisocial impulsivists’’).

Thus, these two groups appear to be quite well individualised.

On the other hand, it is difficult to isolate the specific characteristics of group 1

(‘‘behaviourally conditioned’’), which seems to be intermediate between the other two. Our

Table 2 Results of the logistic regression comparing group 2 (emotionally vulnerable) and group 3 (an-tisocial impulsivists): the contribution of the antisocial-impulsive component

Variables Group 2 % or M (sd) Group 3 OR p

DSM criterion 9 (endangerment) 40.5 % 57.6 % 3.04** 0.009

Accidental discovery of thegambling problem by the family

29.2 % 65.1 % 4.90*** \0.001

TCI ‘‘danger avoidance’’ 57.6 (25.4) 36.7 (19.1) 0.95*** \0.001

TCI ‘‘reward dependence’’ 61.0 (18.0) 53.4 (15.3) 0.97** 0.003

TCI ‘‘persistence’’ 59.6 (27.4) 38.4 (23.6) 0.96*** \0.001

Level of suicide risk1 0.014

Low 58.3 % 51.5 % – –

Moderate 22.9 % 9.1 % 0.27 0.121

High 18.8 % 39.4 % 5.37 0.018

The ORs are given in reference to group 2, i.e. if the OR is greater than 1, the characteristic evaluated isfound more in group 3 than group 2

OR = odds ratio

% = proportions of the procedure studied within the group

M (sd) = mean and standard deviation of the variable studied within the group1 The ‘‘low’’ level is used as reference for the comparison of levels of suicide risk (i.e. the figurescorresponding to ‘‘moderate’’ reflect the comparison of ‘‘moderate’’ versus ‘‘low’’ and those correspondingto ‘‘high’’ reflect the comparison of ‘‘high’’ versus ‘‘low’’)

* p\ 0.05, ** p\ 0.01, *** p\ 0.001

J Gambl Stud

123

results are therefore mixed: descriptively, the groups match our expectations, but the

statistical arguments designed to validate the groups’ solidity are not very strong: the

socio-demographic and personal history data are not sufficient. On the other hand, the

differentiation between these groups is clearly affected by the different types of gambling

practised, and by this fact, the distinction between groups is distinctly established.

These elements force us to highlight a ‘‘functional’’ interpretation of the different

groups of excessive gamblers.

The results of the ‘‘open-ended’’ comparisons between the three groups (‘‘behaviourally

conditioned’’ vs ‘‘emotionally vulnerable’’ and ‘‘emotionally vulnerable’’ vs ‘‘antisocial

impulsivists’’) seem to delineate a particular table for group 3 (‘‘antisocial impulsivists’’):

subjects in this group seem to be distinctly more ‘‘active’’, less sensitive to reward, but

more inclined to take risks. They are more likely to suffer from antisocial personality

disorders (APD). This criterion having been used to set up the group, it cannot be used as a

point of comparison. But, incidentally, it should be noted that APD appears to be an

important indicator of excessive gambling: of the 628 gamblers in the cohort (i.e. 256 non-

problem gamblers and 372 problem gamblers), only 25 met the DSM-IV criteria for

antisocial personality disorder (3.98 %), which is only slightly more than the general

population, according to the D.S.M 3 % of men and 1 % of women (DSM IV-R 2003). But

of these 25 subjects, 21 (84 %) met the criteria for probable pathological gambling (results

not given]. Although these figures are too low to assert a solid statistical correlation, this

impressive percentage acts as a stimulus to research into excessive gambling in subjects

presenting APD.

We therefore have the impression, with group 2 (‘‘emotionally vulnerable’’), of dealing

with a group which validates the hypothesis of gambling as escape, avoidance, self-

medication, and with group 3 (‘‘antisocial impulsivists’’) with another set who use gam-

bling as a stimulant, risk taking, a way of obtaining excitement. On the other hand, in

accordance with the publication by Blaszczynski and Nower (2002), but contrary to that of

Turner et al (2008), there are no arguments to support the contention that group 1 gamblers

are more vulnerable to conditioning. They simply have fewer psychopathological char-

acteristics than the others. The members of this group did not start earlier than the others

and did not have a family history either. Moreover, a study of beliefs and attitudes relative

to gambling, via the GABS questionnaire, tends rather to disprove the idea that these

gamblers have more beliefs or cognitive distortions than the others.

It may be thought that, for most of the items concerned, we are dealing with data which

is part of a continuum, dimensionally rather that categorically. The categories we form

could be an arbitrary classification. This problem occurs with most statistical studies, and

for most of the categories used here (impulsivity, neuroticism, etc. are measured dimen-

sionally). This acts as a reminder of the need for a certain amount of care regarding the

‘‘ontological’’ status of the categories we study: for at least two of them, the ‘‘pathways’’

definitely exist, but this doesn’t mean that they constitute solid diagnostic entities, carved

in stone biologically, for example. In what may be seen as a continuum between ‘‘im-

pulsiveness’’ on the one hand and ‘‘self-medication’’ on the other, we have set up frontiers

to help visualise three separate groups, but there may be intermediate forms between these.

This hypothesis is illustrated in Fig. 1. This diagram which takes our results into account,

leads us to confirm the pertinence of Blaszczynski and Nower’s groups. Indeed, in ac-

cordance with our results, we isolate a group 1 which should be in regression for most of

the time, at psychopathological level. This is an intermediate group and a number of these

gamblers may display characteristics belonging to the other two groups, but to a lesser

extent. We therefore propose a new graphic presentation which is purely descriptive, and

J Gambl Stud

123

which is not intended to incorporate hypothetical etiological data which is genetic or

environmental. This presentation suggests a ‘‘functional’’ interpretation of the ‘‘pathways’’,

according to which what people are looking for in gambling differs according to their

impulsiveness and neuroticism: it therefore clouds Blaszczynski and Nower’s interpreta-

tion according to which group 3 is group 2 ‘‘plus’’ impulsivity, group 2 is group 1 ‘‘plus’’

emotional vulnerability, whereas group 1 is only ‘‘simple’’ conditioning to gambling.

Conclusion

This study has several limits. Firstly, for reasons linked to data collection (begun in 2009,

when online poker was officially forbidden), poker players (‘‘skilful game players’’) are

almost not represented. Moreover, the elements of differentiation between the three groups

descriptively (gender, age, gambling history, etc.) are not very significant. However, our

results tend to support the three pathways model, particularly because ‘‘antisocial impul-

sivist’’ gamblers preferentially choose semi-skilful games (horse or sport betting), whereas

‘‘emotionally vulnerable’’ gamblers are more attracted to games of pure chance (especially

slot machines). The group of ‘‘conditioned gamblers’’ appear to be an intermediate group,

without a different history of gambling, beliefs or cognitive distortions, contrary to the

paper by Turner et al. in (2008).

We therefore support a functional interpretation of gambling addiction: it will affect

impulsive, ‘‘active’’ people on the one hand, who are novelty seekers or transgressors. On

the other hand are subjects who gamble for reasons of ‘‘self-medication’’, for whom

gambling can act as an anti-depressant, a means of escape and avoidance, or a magical

expectation of restoration through fate. The comorbidities which preceded gambling are

Impulsivity, antisocial disorders

Neuroticism, depression, anxiety

Group 3

Group 2

Group 1

Fig. 1 The three ‘‘pathways’’: a functional presentation. This diagram offers a new graphic presentation ofthe three pathways of Blaszczynski and Nower’s (2002). This presentation is purely descriptive and is notintended to incorporate hypothetical etiological data, either genetic or environmental. This presentationsuggests a ‘‘functional’’ interpretation of the ‘‘pathways’’, according to which what people are looking for ingambling differs according to their impulsivity and neuroticism. Group 1 corresponds to ‘‘conditionedgamblers’’, group 2 to ‘‘emotionally vulnerable’’ gamblers and group 3 to ‘‘antisocial impulsivist’’ gamblers,as proposed by Blaszczynski and Nower (2002)

J Gambl Stud

123

mostly involved in the ‘‘self-medication’’ or ‘‘emotionally vulnerable’’ group and are

depression and anxiety.

More targeted research could thus target excessive gambling in people suffering from an

APD on the one hand, and excessive gambling in depressed or anxious people on the other:

these factors should be considered to be factors for vulnerability to excessive gambling.

Moreover, this classification in ‘‘pathways’’ is close to other classifications used for

other addictive behaviour, notably alcoholism. Impulsiveness, self-medication and con-

ditioning can be decisive factors in all kinds of addictive behaviour. We think that other

studies should be performed to explore this ‘‘functional’’ view of pathways for pathological

gambling and also for other types of addiction, with or without drugs.

Acknowledgments We wish to sincerely thank all the staff who contributed to this study (JEU group), fortheir valuable assistance and significant investment. A special thanks to those who collected the data.Members of the JEU Group are: Marie Grall-Bronnec, Gaelle Challet-Bouju, Jean-Luc Venisse, LuciaRomo, Cindy Legauffre, Caroline Dubertret, Irene Codina, Marc Valleur, Marc Auriacombe, Melina Fat-seas, Jean-Marc Alexandre, Pierre-Michel Llorca, Isabelle Chereau-Boudet, Christophe Lancon, DavidMagalon, Michel Reynaud and Mohamed-Ali Gorsane. We also wish to thank Aurelie Wellenstein from theMarmottan Hospital’s resource center for the bibliographic research. This study was supported by both thejoint support of the French Inter-departmental Mission for the fight against drugs and drug addiction(MILDT) and the French National Institute of Health and Medical Research (INSERM), as part of the callfor research projects launched by these two organizations in 2007, and a grant from the French Ministry ofHealth (PHRC 2009–RCB 2008-A01188-47). There were no constraints on publishing. This research wasconducted at the initiative of and coordinated by the Clinical Investigation Unit BALANCED ‘‘BehaviorALAddictioNs and ComplEx mood Disorders’’ of the University Hospital of Nantes, who is the sponsor of thisstudy.

Conflict of interest M.G.B., J.L.V. and G.C.B. declare that the University Hospital of Nantes has receivedfunding from gambling industry (Francaise Des Jeux and Pari Mutuel Urbain) in the form of a sponsorshipwhich supports the gambling section of the BALANCED Unit (the Reference Centre for Excessive Gam-bling). Scientific independence towards gambling industry operators is warranted. MV, IC, CL, DM, MF,ICB, MAG and AG declare that they have no conflicts of interest. There were no constraints on publishing.

References

APA. (2000). DSM-IV-TR. Diagnostic and statistical manual of mental disorders (4th ed text revision ed.).Washington, DC: American Psychiatric Association.

Bayle, F. J., Krebs, M. O., Martin, C., Bouvard, M. P., & Wender, P. (2003). Version francaise de la WenderUtah Rating Scale (WURS). Revue Canadienne de Psychiatrie, 48(2), 132.

Binde, P. (2013). Gambling in Sweden, the cultural and socio-political context. Addiction, 109, 193–198.Blaszczynski, A., McConaghy, N., & Frankova, A. (1990). Boredom proneness in pathological gambling.

Psychological Reports, 67, 35–42.Blaszczynski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling. Addiction,

97, 487–499.Bonnaire, C. (2011). Towards an integrative treatment of pathological gamblers. Encephale, 37(6), 410–417.Bonnaire, C., Bungener, C., Varescon, I. (2009). Subtypes of French pathological gamblers: Comparison of

sensation seeking, alexithymia and depression scores. Journal of Gambling Studies, 28, 455–471.Bouju, G., Hardouin, J. B., Boutin, C., Gorwood, P., Le Bourvellec, J. D., Feuillet, F. (2014). A shorter and

multidimensional version of the gambling attitudes and beliefs survey (GABS-23). Journal of Gam-bling Studies, 30(2), 349–367.

Boutin, C. (2010). Gambling: Luck or strategy? Freely choosing the place of game in your life. Montreal: leseditions de l’homme.

Breen, R. B., & Zuckerman, M. (1999). ‘Chasing’ in gambling behavior: Personality and cognitive deter-minants. Personality Individual Differences, 27, 1097–1111.

Caillois, R. (1957) Les jeux et les hommes, Gallimard, (On Game and men).

J Gambl Stud

123

Chakroun-Vinciguerra, N., Faytout, M., Pelissolo, A., & Swendsen, J. (2005). French validation of a shortversion of the temparament and character inventory (TCI-125). Journal de Therapie Comportementaleet Cognitive, 15(1), 27–33.

Challet-Bouju, G., Hardouin, J. B., Venisse, J. L., Romo, L., Valleur, M., Magalon, D., et al. (2014). Studyprotocol: The JEU cohort study—transversal multiaxial evaluation and 5-year follow-up of a cohort ofFrench gamblers. BMC Psychiatry, 14, 226.

Cloninger, C. R. (1992). The temperament and character inventory-125 (TCI-125; Version 1).Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological model of temperament and

character. Archives of General Psychiatry, 50(12), 975–990.Costes, J. M., Pousset, M., Eroukmanoff, V., Le Nezet, O., Richard, J. B., Guignard, R., Beck, F., Arwidson,

P. (2011). [Rate and practices of gambling in 2010]. Tendances, p 77.Graham, J. R., & Lowenfeld, B. H. (1986). Personality dimensions of the pathological gambler. Journal of

Gambling Behavior, 2, 58–66.Inserm. (2008). Jeux de hasard et d’argent. Inserm: Expertise collective.Jacobs, D. F. (1988). Evidence for a common dissociative like reaction among addicts. Journal of Gambling

Behavior, 4, 27–37.Lecrubier, Y., Sheehan, D. V., Weiller, E., Amorim, P., Bonora, I., Sheehan, K. H., et al. (1997). The mini

international neuropsychiatric interview (MINI). A short diagnostic structured interview: Reliabilityand validity according to the CIDI. European Psychiatry, 12, 224–231.

Ledgerwood, D. M., & Petry, N. M. (2006). Psychological experience of gambling and subtypes ofpathological gamblers. Psychiatry Research, 144, 17–27.

Ledgerwood, D. M., & Petry, N. M. (2010). Subtyping pathological gamblers based on impulsivity, de-pression, and anxiety. Psychology of Addictive Behaviors, 24(4), 680–688.

Lejoyeux, M. (1999). Assesment scales of pathological gambling. Neuropsy, 14, 67–71.Lesieur, H. R. (2001). Cluster analysis of types of inpatient pathological gamblers. Dissertation Abstracts

International, 62B, 2065.Lesieur, H. R., & Blume, S. B. (1987). The South Oaks gambling screen (SOGS): A new instrument for the

identification of pathological gamblers. American Journal of Psychiatry, 144, 1184–1188.Lesieur, H. R., & Blume, S. B. (1991). When lady luck loses: Women and compulsive gambling. In N. Van

Den Bergh (Ed.), Feminist perspectives on addictions (pp. 181–197). New York: Springer PublishingCo.

Martins, S. S., Ghandour, L. A., & Storr, C. L. (2011). Gambling behavior subtypes among respondents withgambling-related problems in a population-based sample. Journal of Addictive Diseases, 30(2),169–180.

McCormick, R. A. (1987). Pathological gambling: A parsimonious need state model. Journal of GamblingBehavior, 3, 257–263.

Milosevic, A., & Ledgerwood, D. M. (2010). The subtyping of pathological gambling: A comprehensivereview. Clinical Psychology Review, 30(8), 988–998.

Moran, E. (1970). Varieties of pathological gambling. The British Journal of Psychiatry, 116, 593–597.Nadeau, L., Valleur, M. (2014). [Pascasius or understanding addictions] (1561), Presses Universitaires de

Montreal.Pelissolo, A., & Lepine, J. (2000). Normative data and factor structure of the temperament and character

inventory (TCI) in the French version. Psychiatry Research, 94, 67–76.Planzer, S., Gray, H. M., & Shaffer, H. J. (2014). Associations between national gambling policies and

disordered gambling prevalence rates within Europe. International Journal of Law and Psychiatry, 37,217–229.

Potenza, M. N. (2006). Should addictive disorders include non-substance-related conditions? Addiction,101(Suppl 1), 142–151.

Shaffer, H. J., & Hall, M. N. (2001). Updating and refining prevalence estimates of disordered gamblingbehaviour in the United States and Canada. Canadian Journal of Public Health, 92, 168–172.

Sharpe, L. (2002). A reformulated cognitive-behavioural model of problem gambling: A biopsychosocialperspective. Clinical Psychology Review, 22, 1–25.

Sproston, K., Erens, B., & Orford, J. (2000). Gambling behaviour in Britain: Results from the gamblingprevalence survey. London: National Center for Social Research.

Steel, Z., & Blaszczynski, A. (1996). The factorial structure of pathological gambling. Journal of GamblingStudies, 12, 3–20.

Stewart, S. H., & Zack, M. (2008). Development and psychometric evaluation of a three-dimensionalgambling motives questionnaire. Addiction, 103, 1110–1117.

Toneatto, T., & Millar, G. (2004). Assessing and treating problem gambling: Empirical status and promisingtrends. Canadian Journal of Psychiatry, 49, 517–525.

J Gambl Stud

123

Turner, N. E., Jain, U., Spence, W., & Zangeneh, M. (2008). Pathways to pathological gambling: Com-ponent analysis of variables related to pathological gambling. International Gambling Studies, 8,281–298.

Vachon, D. D., & Bagby, R. M. (2009). Pathological gambling subtypes. Psychological Assessment, 21,608–615.

Valleur, M., Bucher, C. (2006). [Pathological gambling] Armand Colin.Ward, M. F., Wender, P. H., & Reimherr, F. W. (1993). The wender utah rating scale: An aid in the

retrospective diagnosis of childhood attention deficit hyperactivity disorder. American Journal ofPsychiatry, 150, 885–890.

Wardle, H., Moody, A., Spence, S., Orford, J., Volberg, R., & Jotangia, D. (2011). British gamblingprevalence survey 2010. London: National Center for Social Research.

Wardle, H., Sproston, K., Orford, J., Erens, B., Griffiths, M., & Constantine, R. (2007). British gamblingprevalence survey 2007. London: National Center for Social Research.

J Gambl Stud

123